Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x16340898390>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x16340958400>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.2.1
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    print('image_width, image_height, image_channels',image_width, image_height, image_channels)
    input_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name = 'input_real')
    print('z_dim',z_dim)
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name = 'input_z')
    learning_rate = tf.placeholder(tf.float32, [], name = 'input_lr')

    return input_real, inputs_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

image_width, image_height, image_channels 28 28 3
z_dim 100
Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

# Constant for the leaky relu.
alpha = 0.2

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    with tf.variable_scope('discriminator', reuse=reuse):
        # Input layer is 32x32x3    28x28xOUT
        x1 = tf.layers.conv2d(images, 64, 5, strides = 2, padding = 'same')
        relu1 = tf.maximum(alpha*x1, x1)
        # 16x16x64  14x14xOUT
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides = 2, padding = 'same')
        bn2 = tf.layers.batch_normalization(x2, training = True)
        relu2 = tf.maximum(alpha*bn2, bn2)
        # 8x8x128   7x7xOUT
        
        # x3 = tf.layers.conv2d(relu2, 256, 5, strides = 2, padding = 'same')
        # bn3 = tf.layers.batch_normalization(x3, training = True)
        # relu3 = tf.maximum(alpha*bn3, bn3)
        # 4x4x256
        
        flat = tf.reshape(relu2, (-1, 7*7*128) )
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
    return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    with tf.variable_scope('generator', reuse = not is_train):
        # First fully connected layer
        x = tf.layers.dense(z, 7*7*256)
        # Reshape it to start the convolution stack
        x1 = tf.reshape(x, (-1, 7, 7, 256) )
        x1 = tf.layers.batch_normalization(x1, training = is_train)
        x1 = tf.maximum(alpha*x1, x1)
        # 4x4x512 now      7x7xOUT
        
        x2 = tf.layers.conv2d_transpose(x1, 128, 5, strides = 2, padding = 'same')
        x2 = tf.layers.batch_normalization(x2, training = is_train)
        x2 = tf.maximum(alpha*x2, x2)
        # 8x8x256 now     14x14xOUT   
        
        # x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides = 2, padding = 'same')
        # x3 = tf.layers.batch_normalization(x3, training = is_train)
        # x3 = tf.maximum(alpha*x3, x3)
        # 16x16x128     28x28xOUT
     
        # Output layer, 32x32x3  28x28xOUT
        logits = tf.layers.conv2d_transpose(x2, out_channel_dim, 5, strides = 2, padding = 'same')
        
        out = tf.tanh(logits)
        
    return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # Get weights and bias to update
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

import pickle as pkl
def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # tf.reset_default_graph()
    print('*data_shape[1:4]',*data_shape[1:4])
    print('z_dim',z_dim)
    input_real, input_z, input_lr = model_inputs(*data_shape[1:4], z_dim)
    out_channel_dim = data_shape[3]
    d_loss, g_loss = model_loss(input_real, input_z, out_channel_dim)
 
    d_train_opt, g_train_opt = model_opt(d_loss, g_loss, input_lr, beta1)
                
    saver = tf.train.Saver()
    sample_z = np.random.uniform(-1, 1, size=(72, z_dim))
    
    samples, losses = [], []
    steps = 0
    print_every = 10
    show_every = 100
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
        
            print('epoch_i', epoch_i)
            for batch_images in get_batches(batch_size):
                steps += 1

                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))

                # Run optimizers
                _ = sess.run(d_train_opt, feed_dict={input_real: batch_images, input_z: batch_z, input_lr: learning_rate})
                _ = sess.run(g_train_opt, feed_dict={input_real: batch_images, input_z: batch_z, input_lr: learning_rate})

                if steps % print_every == 0:
                    # At the end of each epoch, get the losses and print them out
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    # Save losses to view after training
                    losses.append((train_loss_d, train_loss_g))

                if steps % show_every == 0:
                    gen_samples = sess.run(
                                   generator(input_z, out_channel_dim, is_train=False),
                                   feed_dict={input_z: sample_z})
                    samples.append(gen_samples)

                    show_generator_output(sess, 5, input_z, out_channel_dim, data_image_mode)

        saver.save(sess, './checkpoints/generator.ckpt')

    with open('samples.pkl', 'wb') as f:
        pkl.dump(samples, f)
    
    return losses, samples

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 64
z_dim = 10
learning_rate = 0.0005
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

*data_shape[1:4] 28 28 1
z_dim 10
image_width, image_height, image_channels 28 28 1
z_dim 10
epoch_i 0
Epoch 1/2... Discriminator Loss: 2.1180... Generator Loss: 0.4046
Epoch 1/2... Discriminator Loss: 1.9464... Generator Loss: 0.4755
Epoch 1/2... Discriminator Loss: 1.3302... Generator Loss: 0.7492
Epoch 1/2... Discriminator Loss: 1.3453... Generator Loss: 0.5141
Epoch 1/2... Discriminator Loss: 0.9035... Generator Loss: 1.2886
Epoch 1/2... Discriminator Loss: 1.2338... Generator Loss: 0.5764
Epoch 1/2... Discriminator Loss: 1.5255... Generator Loss: 4.0629
Epoch 1/2... Discriminator Loss: 1.0512... Generator Loss: 0.9112
Epoch 1/2... Discriminator Loss: 0.5561... Generator Loss: 2.0744
Epoch 1/2... Discriminator Loss: 1.0644... Generator Loss: 1.4419

Epoch 1/2... Discriminator Loss: 0.5713... Generator Loss: 2.0058
Epoch 1/2... Discriminator Loss: 0.8372... Generator Loss: 0.8424
Epoch 1/2... Discriminator Loss: 0.7605... Generator Loss: 1.0242
Epoch 1/2... Discriminator Loss: 0.6015... Generator Loss: 1.7645
Epoch 1/2... Discriminator Loss: 1.0794... Generator Loss: 0.5701
Epoch 1/2... Discriminator Loss: 0.9590... Generator Loss: 1.2370
Epoch 1/2... Discriminator Loss: 0.8611... Generator Loss: 1.1771
Epoch 1/2... Discriminator Loss: 1.0625... Generator Loss: 0.5711
Epoch 1/2... Discriminator Loss: 0.6924... Generator Loss: 1.0959
Epoch 1/2... Discriminator Loss: 0.4872... Generator Loss: 2.0534

Epoch 1/2... Discriminator Loss: 0.6153... Generator Loss: 1.4669
Epoch 1/2... Discriminator Loss: 0.5705... Generator Loss: 1.6273
Epoch 1/2... Discriminator Loss: 0.6093... Generator Loss: 1.1803
Epoch 1/2... Discriminator Loss: 0.5151... Generator Loss: 1.8548
Epoch 1/2... Discriminator Loss: 0.3045... Generator Loss: 2.9212
Epoch 1/2... Discriminator Loss: 1.4056... Generator Loss: 0.3446
Epoch 1/2... Discriminator Loss: 0.3382... Generator Loss: 2.3063
Epoch 1/2... Discriminator Loss: 0.6738... Generator Loss: 2.9013
Epoch 1/2... Discriminator Loss: 1.2490... Generator Loss: 0.4156
Epoch 1/2... Discriminator Loss: 0.7262... Generator Loss: 1.0126

Epoch 1/2... Discriminator Loss: 0.4110... Generator Loss: 1.8891
Epoch 1/2... Discriminator Loss: 0.7495... Generator Loss: 5.7261
Epoch 1/2... Discriminator Loss: 0.6516... Generator Loss: 0.9629
Epoch 1/2... Discriminator Loss: 0.3245... Generator Loss: 2.1813
Epoch 1/2... Discriminator Loss: 0.2302... Generator Loss: 3.2905
Epoch 1/2... Discriminator Loss: 0.3631... Generator Loss: 2.8545
Epoch 1/2... Discriminator Loss: 0.3706... Generator Loss: 2.0940
Epoch 1/2... Discriminator Loss: 0.3693... Generator Loss: 1.8259
Epoch 1/2... Discriminator Loss: 0.4108... Generator Loss: 1.5695
Epoch 1/2... Discriminator Loss: 1.0456... Generator Loss: 0.5555

Epoch 1/2... Discriminator Loss: 0.4371... Generator Loss: 3.5555
Epoch 1/2... Discriminator Loss: 0.2951... Generator Loss: 2.1190
Epoch 1/2... Discriminator Loss: 0.8201... Generator Loss: 0.7331
Epoch 1/2... Discriminator Loss: 0.6413... Generator Loss: 1.2268
Epoch 1/2... Discriminator Loss: 0.2317... Generator Loss: 2.5900
Epoch 1/2... Discriminator Loss: 0.2532... Generator Loss: 4.2592
Epoch 1/2... Discriminator Loss: 0.3373... Generator Loss: 3.1641
Epoch 1/2... Discriminator Loss: 0.2347... Generator Loss: 3.0794
Epoch 1/2... Discriminator Loss: 1.2466... Generator Loss: 0.4309
Epoch 1/2... Discriminator Loss: 0.2038... Generator Loss: 2.9134

Epoch 1/2... Discriminator Loss: 0.4898... Generator Loss: 1.1745
Epoch 1/2... Discriminator Loss: 0.1883... Generator Loss: 2.5138
Epoch 1/2... Discriminator Loss: 0.3474... Generator Loss: 2.1869
Epoch 1/2... Discriminator Loss: 0.2805... Generator Loss: 1.9656
Epoch 1/2... Discriminator Loss: 0.2190... Generator Loss: 2.3147
Epoch 1/2... Discriminator Loss: 0.1973... Generator Loss: 3.8676
Epoch 1/2... Discriminator Loss: 0.2347... Generator Loss: 2.9488
Epoch 1/2... Discriminator Loss: 0.2572... Generator Loss: 1.7799
Epoch 1/2... Discriminator Loss: 0.2878... Generator Loss: 1.6852
Epoch 1/2... Discriminator Loss: 0.8429... Generator Loss: 0.6476

Epoch 1/2... Discriminator Loss: 0.3412... Generator Loss: 1.7787
Epoch 1/2... Discriminator Loss: 0.2516... Generator Loss: 1.9967
Epoch 1/2... Discriminator Loss: 0.2292... Generator Loss: 2.0737
Epoch 1/2... Discriminator Loss: 0.0706... Generator Loss: 4.6781
Epoch 1/2... Discriminator Loss: 0.1098... Generator Loss: 2.8819
Epoch 1/2... Discriminator Loss: 0.3274... Generator Loss: 1.8860
Epoch 1/2... Discriminator Loss: 0.2084... Generator Loss: 3.1486
Epoch 1/2... Discriminator Loss: 0.4413... Generator Loss: 4.0153
Epoch 1/2... Discriminator Loss: 0.1001... Generator Loss: 3.3394
Epoch 1/2... Discriminator Loss: 0.1499... Generator Loss: 4.2828

Epoch 1/2... Discriminator Loss: 0.1134... Generator Loss: 4.3384
Epoch 1/2... Discriminator Loss: 0.1712... Generator Loss: 2.4412
Epoch 1/2... Discriminator Loss: 0.9540... Generator Loss: 4.6023
Epoch 1/2... Discriminator Loss: 0.4654... Generator Loss: 1.7775
Epoch 1/2... Discriminator Loss: 0.4028... Generator Loss: 2.4416
Epoch 1/2... Discriminator Loss: 0.1644... Generator Loss: 2.6804
Epoch 1/2... Discriminator Loss: 0.1466... Generator Loss: 2.8807
Epoch 1/2... Discriminator Loss: 0.2012... Generator Loss: 2.4888
Epoch 1/2... Discriminator Loss: 0.1530... Generator Loss: 3.9742
Epoch 1/2... Discriminator Loss: 0.3215... Generator Loss: 2.3065

Epoch 1/2... Discriminator Loss: 0.3378... Generator Loss: 2.3330
Epoch 1/2... Discriminator Loss: 0.2833... Generator Loss: 1.9182
Epoch 1/2... Discriminator Loss: 0.1967... Generator Loss: 2.3647
Epoch 1/2... Discriminator Loss: 0.2082... Generator Loss: 4.2074
Epoch 1/2... Discriminator Loss: 0.1453... Generator Loss: 2.7367
Epoch 1/2... Discriminator Loss: 0.1765... Generator Loss: 2.4883
Epoch 1/2... Discriminator Loss: 0.1920... Generator Loss: 2.4536
Epoch 1/2... Discriminator Loss: 0.1518... Generator Loss: 2.5655
Epoch 1/2... Discriminator Loss: 0.1188... Generator Loss: 3.2504
Epoch 1/2... Discriminator Loss: 0.1064... Generator Loss: 3.0561

Epoch 1/2... Discriminator Loss: 0.5267... Generator Loss: 1.0740
Epoch 1/2... Discriminator Loss: 1.2156... Generator Loss: 0.5784
Epoch 1/2... Discriminator Loss: 0.5282... Generator Loss: 1.5498
epoch_i 1
Epoch 2/2... Discriminator Loss: 0.4393... Generator Loss: 2.2918
Epoch 2/2... Discriminator Loss: 0.3307... Generator Loss: 2.1136
Epoch 2/2... Discriminator Loss: 0.4740... Generator Loss: 1.7117
Epoch 2/2... Discriminator Loss: 0.2866... Generator Loss: 2.5325
Epoch 2/2... Discriminator Loss: 0.6643... Generator Loss: 1.8096
Epoch 2/2... Discriminator Loss: 0.3966... Generator Loss: 1.8050
Epoch 2/2... Discriminator Loss: 0.4007... Generator Loss: 1.5133

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 128
z_dim = 100
learning_rate = 0.0005
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 30

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

*data_shape[1:4] 28 28 3
z_dim 100
image_width, image_height, image_channels 28 28 3
z_dim 100
epoch_i 0
Epoch 1/30... Discriminator Loss: 2.3496... Generator Loss: 0.2526
Epoch 1/30... Discriminator Loss: 1.1943... Generator Loss: 0.9957
Epoch 1/30... Discriminator Loss: 1.1379... Generator Loss: 1.7369
Epoch 1/30... Discriminator Loss: 0.8913... Generator Loss: 1.0148
Epoch 1/30... Discriminator Loss: 0.3333... Generator Loss: 2.9946
Epoch 1/30... Discriminator Loss: 0.2026... Generator Loss: 2.4891
Epoch 1/30... Discriminator Loss: 0.1069... Generator Loss: 3.0749
Epoch 1/30... Discriminator Loss: 0.2437... Generator Loss: 8.3053
Epoch 1/30... Discriminator Loss: 0.2604... Generator Loss: 1.6914
Epoch 1/30... Discriminator Loss: 0.1176... Generator Loss: 5.7855

Epoch 1/30... Discriminator Loss: 0.3604... Generator Loss: 1.3752
Epoch 1/30... Discriminator Loss: 1.2467... Generator Loss: 0.4168
Epoch 1/30... Discriminator Loss: 0.2022... Generator Loss: 4.4801
Epoch 1/30... Discriminator Loss: 0.8880... Generator Loss: 0.7719
Epoch 1/30... Discriminator Loss: 2.2553... Generator Loss: 0.1446
Epoch 1/30... Discriminator Loss: 0.1473... Generator Loss: 11.7457
Epoch 1/30... Discriminator Loss: 0.2671... Generator Loss: 3.1564
Epoch 1/30... Discriminator Loss: 0.2016... Generator Loss: 2.0625
Epoch 1/30... Discriminator Loss: 0.5013... Generator Loss: 8.6088
Epoch 1/30... Discriminator Loss: 0.2607... Generator Loss: 2.3420

Epoch 1/30... Discriminator Loss: 1.1323... Generator Loss: 0.4821
Epoch 1/30... Discriminator Loss: 0.8395... Generator Loss: 0.6650
Epoch 1/30... Discriminator Loss: 0.5399... Generator Loss: 1.6871
Epoch 1/30... Discriminator Loss: 0.3746... Generator Loss: 2.0340
Epoch 1/30... Discriminator Loss: 0.9041... Generator Loss: 0.7041
Epoch 1/30... Discriminator Loss: 0.1936... Generator Loss: 2.3749
Epoch 1/30... Discriminator Loss: 0.0571... Generator Loss: 6.2721
Epoch 1/30... Discriminator Loss: 1.2374... Generator Loss: 0.4244
Epoch 1/30... Discriminator Loss: 0.5098... Generator Loss: 1.5964
Epoch 1/30... Discriminator Loss: 1.0111... Generator Loss: 1.3180

Epoch 1/30... Discriminator Loss: 1.0592... Generator Loss: 0.5584
Epoch 1/30... Discriminator Loss: 0.3259... Generator Loss: 2.0204
Epoch 1/30... Discriminator Loss: 1.5010... Generator Loss: 4.2197
Epoch 1/30... Discriminator Loss: 0.6566... Generator Loss: 2.0959
Epoch 1/30... Discriminator Loss: 1.6224... Generator Loss: 0.2752
Epoch 1/30... Discriminator Loss: 0.4089... Generator Loss: 2.0430
Epoch 1/30... Discriminator Loss: 0.7356... Generator Loss: 0.9529
Epoch 1/30... Discriminator Loss: 0.2301... Generator Loss: 3.5037
Epoch 1/30... Discriminator Loss: 0.2501... Generator Loss: 6.4610
Epoch 1/30... Discriminator Loss: 1.7335... Generator Loss: 0.2456

Epoch 1/30... Discriminator Loss: 0.1264... Generator Loss: 5.3341
Epoch 1/30... Discriminator Loss: 0.4313... Generator Loss: 1.5480
Epoch 1/30... Discriminator Loss: 0.2882... Generator Loss: 2.4762
Epoch 1/30... Discriminator Loss: 0.9933... Generator Loss: 0.6183
Epoch 1/30... Discriminator Loss: 0.5412... Generator Loss: 1.5208
Epoch 1/30... Discriminator Loss: 0.4129... Generator Loss: 2.0909
Epoch 1/30... Discriminator Loss: 0.1917... Generator Loss: 3.6352
Epoch 1/30... Discriminator Loss: 0.1559... Generator Loss: 4.1571
Epoch 1/30... Discriminator Loss: 0.2595... Generator Loss: 2.8662
Epoch 1/30... Discriminator Loss: 0.4263... Generator Loss: 6.3892

Epoch 1/30... Discriminator Loss: 0.3028... Generator Loss: 4.6209
Epoch 1/30... Discriminator Loss: 0.1233... Generator Loss: 5.4055
Epoch 1/30... Discriminator Loss: 0.1694... Generator Loss: 6.5270
Epoch 1/30... Discriminator Loss: 0.3698... Generator Loss: 2.4868
Epoch 1/30... Discriminator Loss: 0.1096... Generator Loss: 4.5425
Epoch 1/30... Discriminator Loss: 0.2410... Generator Loss: 3.0908
Epoch 1/30... Discriminator Loss: 0.5699... Generator Loss: 1.4615
Epoch 1/30... Discriminator Loss: 0.2345... Generator Loss: 3.6579
Epoch 1/30... Discriminator Loss: 0.1647... Generator Loss: 6.4599
Epoch 1/30... Discriminator Loss: 0.1994... Generator Loss: 5.6919

Epoch 1/30... Discriminator Loss: 0.6523... Generator Loss: 6.0654
Epoch 1/30... Discriminator Loss: 1.0159... Generator Loss: 0.7166
Epoch 1/30... Discriminator Loss: 0.1560... Generator Loss: 5.7679
Epoch 1/30... Discriminator Loss: 0.1834... Generator Loss: 7.4728
Epoch 1/30... Discriminator Loss: 0.2543... Generator Loss: 5.0645
Epoch 1/30... Discriminator Loss: 0.1503... Generator Loss: 5.7628
Epoch 1/30... Discriminator Loss: 0.9896... Generator Loss: 0.7981
Epoch 1/30... Discriminator Loss: 0.1103... Generator Loss: 6.7093
Epoch 1/30... Discriminator Loss: 0.3232... Generator Loss: 4.5420
Epoch 1/30... Discriminator Loss: 0.3139... Generator Loss: 3.0311

Epoch 1/30... Discriminator Loss: 0.7455... Generator Loss: 1.3249
Epoch 1/30... Discriminator Loss: 0.0550... Generator Loss: 6.3798
Epoch 1/30... Discriminator Loss: 0.1965... Generator Loss: 6.0400
Epoch 1/30... Discriminator Loss: 0.1665... Generator Loss: 9.4722
Epoch 1/30... Discriminator Loss: 0.9785... Generator Loss: 14.4287
Epoch 1/30... Discriminator Loss: 0.2562... Generator Loss: 8.4339
Epoch 1/30... Discriminator Loss: 0.2162... Generator Loss: 2.7654
Epoch 1/30... Discriminator Loss: 0.4589... Generator Loss: 9.0235
Epoch 1/30... Discriminator Loss: 0.7618... Generator Loss: 1.9900
Epoch 1/30... Discriminator Loss: 0.2462... Generator Loss: 5.3641

Epoch 1/30... Discriminator Loss: 1.1192... Generator Loss: 2.0172
Epoch 1/30... Discriminator Loss: 0.1905... Generator Loss: 5.1488
Epoch 1/30... Discriminator Loss: 0.4943... Generator Loss: 7.0537
Epoch 1/30... Discriminator Loss: 0.2474... Generator Loss: 4.1115
Epoch 1/30... Discriminator Loss: 0.3481... Generator Loss: 2.3558
Epoch 1/30... Discriminator Loss: 1.4068... Generator Loss: 12.9333
Epoch 1/30... Discriminator Loss: 0.1588... Generator Loss: 6.3236
Epoch 1/30... Discriminator Loss: 0.1303... Generator Loss: 6.6784
Epoch 1/30... Discriminator Loss: 0.9308... Generator Loss: 9.5450
Epoch 1/30... Discriminator Loss: 0.2685... Generator Loss: 3.8490

Epoch 1/30... Discriminator Loss: 0.4005... Generator Loss: 3.5125
Epoch 1/30... Discriminator Loss: 0.2102... Generator Loss: 4.4440
Epoch 1/30... Discriminator Loss: 0.7897... Generator Loss: 0.9460
Epoch 1/30... Discriminator Loss: 0.1339... Generator Loss: 7.1547
Epoch 1/30... Discriminator Loss: 0.0975... Generator Loss: 9.1087
Epoch 1/30... Discriminator Loss: 0.1345... Generator Loss: 7.6104
Epoch 1/30... Discriminator Loss: 0.3988... Generator Loss: 2.1743
Epoch 1/30... Discriminator Loss: 0.5625... Generator Loss: 2.3821
Epoch 1/30... Discriminator Loss: 0.3650... Generator Loss: 9.2626
Epoch 1/30... Discriminator Loss: 0.2662... Generator Loss: 4.1816

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.